Median Prices Still Crashing in Some Neighborhoods

As we were digging through the February data from the NWMLS, some of us noticed something interesting. While the county-wide median price was basically flat year-over-year, there is a pretty strong divergence from region to region around the county.

Here’s what the YOY median price breakdown looks like at the region level:

SW King: -12.8%
SE King: -11.1%
Seattle: +1.4%
N. King: -17.2%
Eastside: +1.0%

See what I mean? Outside of Seattle proper and the Eastside, median prices still seem to be falling quite fast. Eric Pryne at the Seattle Times noticed this as well, making it the focus of his report last week.

In order to explore this further, I have taken the February data from the NWMLS and plugged it into a Tableau map. Unfortunately, Tableau does not let you draw arbitrary boundary lines, so for the full map of the NWMLS areas with the boundary lines drawn, you’ll have to go here.

Hit the jump for the interactive map.

With just three and six closed sales, Vashon and Mercer Islands saw the most extreme swings in their respective median prices, falling 38% on Vashon and rising a whopping 124% on Mercer. Most of the other neighborhoods around King County fell within a more normal range, with the exception of the Capitol Hill area, where the median shot up over 45%. That could be because last year that area saw just 17 sales in February.

Ten of thirty areas saw YOY gains in their median prices. The average gain among those ten was 22% (skewed high by Mercer), and the median gain was 6.4%.

Of the twenty areas that saw YOY declines in their median price, the average loss was -12.2%, and the median was -13.2%. In other words, in the neighborhoods where prices are still falling, they tend to still be falling pretty fast.


About The Tim

Tim Ellis is the founder of Seattle Bubble. His background in engineering and computer / internet technology, a fondness of data-based analysis of problems, and an addiction to spreadsheets all influence his perspective on the Seattle-area real estate market.

49 comments:

  1. 1

    Even the Green Uptick Numbers Look Bogus

    Notice, they’re from the Eastside and Seattle areas; higher priced Pink Pony heaven locations. More wealthy sellers too IMO, IOWs they can afford to list a property for a year or two without going bankrupt….so refuse to sell sweetheart deals like the more blue collar locations.

    It’s like the price vultures patiently waiting in the trees for the seller cattle to finally collapse, eventually the economic realities hit the rich fatter cattle too.

  2. 2
    skibum1175 says:

    Nice illustration of how there’re huge swings and variability in the market.

  3. 3
    Dave0 says:

    I have a suspicion that Capitol Hill is holding up better than most due to the anticipation of the light rail station that is being built there.

  4. 4
    The Tim says:

    RE: Dave0 @ 3 – I think the main reason it’s so dramatic is just a quirk of the noisy nature of the median when you start looking at smaller sample sizes. Here’s the SFH median for Capitol Hill (area 390) since 2007:

    February 2009 was an abnormally low point, and February 2010 seems to be abnormally high, thus the dramatic 45% gain.

  5. 5
    ARDELL says:

    Small geographic area and one month’s worth of sales never works well. I’ll try it with a 90 day sample on a rolling basis.

    Also “area 600” as being “Eastside” is a bit off, given area 600 incorporates a huge area that goes way past what we call “Eastside”.

    I’ll try to do a companion post on RCG with some variances. I don’t usually use mls “area” numbers as I am in favor of them being removed and replaced with “real” neighborhoods or at least zip codes. Something more in line with what consumers generally would call those areas, and not some “secret code” system that needs a decoder ring for members only.

  6. 6
    The Tim says:

    By ARDELL @ 5:

    Also “area 600” as being “Eastside” is a bit off, given area 600 incorporates a huge area that goes way past what we call “Eastside”.

    I am just going by the way the NWMLS groups the areas in their King County Breakouts pdf.

    I don’t usually use mls “area” numbers as I am in favor of them being removed and replaced with “real” neighborhoods or at least zip codes. Something more in line with what consumers generally would call those areas, and not some “secret code” system that needs a decoder ring for members only.

    I agree with that for sure. I only use the NWMLS areas because that’s the only way the data is published. If I had a reliable way to consistently acquire data sorted by zip code instead I’d use that in a heartbeat. Of course I’d also need years worth of historical data to make anything useful out of it, so that would be a bit of a hangup if they just suddenly decided to switch.

  7. 7
    East Side Real Estate Agent says:

    Those graphs and numbers dont reflect sales that were not officially reported. Things are much better than they want you to think.

  8. 8
    ARDELL says:

    Tim,

    Since I can do it without area codes, and have historical data as well, I won’t use the areas. I’ll see what I can do to complement your presentation. I’m curious because the median YOY is not down (last I looked a few days ago) so it seems “off” to me for so many of those areas to be down and some by a lot, and very few compensating “ups”. If that is true, how can the median not be down? I’ll play with it.

  9. 9
    The Tim says:

    By ARDELL @ 8:

    I’m curious because the median YOY is not down (last I looked a few days ago) so it seems “off” to me for so many of those areas to be down and some by a lot, and very few compensation “ups”. If that is true, how can the median not be down?

    The answer to that question is easy, and I have addressed it here a couple of times in the past. Here’s the most recent post on that subject from July ’09: Median Price Still Being Distorted by Geographic Shifts in Sales

  10. 10
    ARDELL says:

    East Side Real Estate Agent,

    Your comment looks like a red herring purporting that there is a ton of “better” sales off the mls. I have recently tested that theory, and it is not the case. There are some sales outside the mls, not many, and those are more likely to be worse (short sales and foreclosures) than better.

    Where is all the magical “better” data that you are suggesting exists outside of the data Tim is using?

  11. 11
    ARDELL says:

    Tim,

    Help me with this one. Say 5 people buy homes for $500,000 to $600,000 and those are the only sales in a month. Next month 5 people buy homes for $300,000 to $400,000 and those are the only sales. Wouldn’t looking at “medians” erroneously suggest prices down, when it has nothing at all to do with prices down, and just different priced homes “in the mix”? Doesn’t chasing medians merely reflect the change in which buyers happen to be out that month, and are not a reflection of prices going up or down per house at all? RE: The Tim @ 9

  12. 12
    Hugh Dominic says:

    RE: ARDELL @ 10 – ESREA is a known troll. Disregard.

  13. 13
    David Losh says:

    RE: ARDELL @ 11

    Doesn’t this same reasoning apply to your bottom call?

  14. 14
    The Tim says:

    RE: ARDELL @ 11 – Well yes, that is a known problem with medians, and the reason that the Case-Shiller index is the preferred measure for actual home price changes.

    However, I was pointing out that it is easy to see how all of the individual areas could have declining median prices while the county-wide median shows an increase (or other similar scenarios). Just mix up the areas where the sales are taking place.

  15. 15

    RE: The Tim @ 14 – To some extent it’s also just how the areas are grouped. If you look at the individual areas (all 29-30 of them) there are roughly the same number of areas that are down less than 25k as up less than 25k, as up more than 25k. However, there are about 2.5x as many that are down more than 25k.

    Then you have to look at the number of sales in each area. There are 8 areas with more than 50 sales, and of those 2 are down more than 25k, 2 are down less than 25k, 3 are up less than 25k and 2 is up more than 25k. So if you look at those higher volume areas, it’s roughly even.

    Finally, as you note the mix is a lot different. One area when from 17 sales to 70 sales, which is a huge difference, and it’s an area where the median was over $500,000 both years.

  16. 16
    Steve Tytler says:

    Great graphs, as usual.

    This clearly shows what I mean when I say there is really no such thing as the “Seattle Housing Market” which is very broadly defined by Case-Schiller and other indices. Home price appreciation/depreciation varies so dramatically from neighborhood to neighborhood, and city to city, that putting a single number on the whole area is practically meaningless.

    That’s why predicting home prices on a region-wide basis is very difficult. If you’re in Puyallup you’re wondering when the pain will ever end, while if you’re in the desirable areas of Seattle and the Eastside you’re thinking the worst may be over. That’s why I always make my market predictions in a range rather than a specific number.

    It’s good news to see some areas moving into the green. As usual, the outlying areas will lag neighborhoods close to the major urban job centers. This could be the beginning of the bottom some of us have been waiting for — but it will be at least another year before we can know for sure because as you all know, you can only truly call a “bottom” in retrospect.

    Overall, I’m optimistic for Seattle and the Eastside this year.

  17. 17
    laterite says:

    Is it possible to run this for Snohomish County as well? I am just now stepping into the Tableau sandbox so I might just hack away at it on my own too. Thanks.

  18. 18
    ARDELL says:

    One of the things I have tried is to segregate the majority of sales by price and then check that as to up or down. Do you think that improves accuracy? I am not criticizing your methods, I’m attempting to collaborate, given I do have access to the data. Maybe we can “play” in the mls together some day and come up with a more valid solution. Your thoughts are appreciated. RE: The Tim @ 14

  19. 19
    ARDELL says:

    No, I don’t think so, David. I think that there were likely more high end than low end at that time, given the high end people were not waiting on the credit to be announced. So sales were more likely a combination of investors and high end vs. skewed to the lower priced sales by owner occupant buyers back in February/March of 2009. RE: David Losh @ 13

  20. 20
    David Losh says:

    RE: ARDELL @ 19

    The median price which you used for your bottom call is statistical data. You can make statistics do what ever you want without qualifying the base.

    Steve Tytler coincidently is here today with his own bottom calling, he’s selling mortgages.

  21. 21
    ARDELL says:

    David,

    The “statistical data” is not something I “make do what I want” when we are talking median price for King County sales. Any and every agent with access to the mls can check that number both when I did it and now in hindsight. It was not “manipulated” as you suggest. It is the straight up number you get if you simply put in SOLD, March 1 to March 31 2009, Residential, King County and then hit the “statistics” button.

    I did not “monkey” with the data as you seem to be suggesting.

  22. 22
    ARDELL says:

    David,

    The “statistical data” is not something I “make do what I want” when we are talking median price for King County sales. Any and every agent with access to the mls can check that number both when I did it and now in hindsight. It was not “manipulated” as you suggest. It is the straight up number you get if you simply put in SOLD, March 1 to March 31 2009, Residential, King County and then hit the “statistics” button.

    I did not “monkey” with the data as you seem to be suggesting.

    For the record…I have a “new” bottom call as well :) I think the market will be lower in 4th quarter 2010 than it was back in March 2009, the previous and still standing “bottom”.

  23. 23
    Hugh Dominic says:

    I misread the headline. I thought it said, “Prices stable to rising in some Seattle neighborhoods.”

  24. 24
    ARDELL says:

    North may or may not be up…but South is definitely down and down again :) and we ain’t talking about below the dixie line. RE: Hugh Dominic @ 23

  25. 25
    David Losh says:

    RE: ARDELL @ 22

    What I’m saying is that the real estate market has been frozen by a massive infusion of government cash. It was an easy call.

    The problem buyers will have is that once the government stops it’s intrusion the market will continue it’s decline.

  26. 26
    Hugh Dominic says:

    By David Losh @ 25:

    RE: once the government stops it’s intrusion the market will continue it’s decline.

    …at which point the goverment will resume it’s intrusion. Further declines of any significance will not be permitted. That is the policy for this administration. Get used to it.

  27. 27
    Hugh Dominic says:

    RE: Steve Tytler @ 16 – there’s something fishy to me about the RE slogan that “all real estate is (hyper-)local.” It just seems like the agent is trying to get a buyer to disregard major systemic downward price pressure. Like, they have to acknowledge the sweeping declines but maybe this one house you’re looking at is immune.

    I call shenanigans.

  28. 28
    Daniel says:

    RE: The Tim @ 4 – Is there any large scale statistical data on property values (preferrably by geographic region) from which one can infer something about the distribution (and how strongly it is skewed from normal)?

    If this exists one could get a good estimate of the variance of the median by drawing replicas with the typical size of a month of data from those distributions and calculating the median for each replica. One then can calculate the variance of this to obtain a kind of Monte Carlo estimate. The goal of this would be qualitative, getting some sort of an intuition what kind of fluctuations are statistically significant.

    Darn it: I read this blog for recreation and all I think about is work =)

  29. 29
    shawn says:

    RE: Steve Tytler @ 16 – you make a lot of sense, and I like your posts, but for me it is not nice to see areas getting into the green. I still cannot understand why RE folks see things as getting better when I have to pay more for a home? Call me silly, but I see things getting better when I can pay less.

  30. 30
    buystocks says:

    I went to several open houses over the weekend. One real estate agent told me we are at the bottom, and another one told me that there is now a housing shortage… So, there you have it, we have hit bottom. (BTW, open house traffic was very quiet)

  31. 31
    David Losh says:

    RE: Hugh Dominic @ 26

    Though I share that concern there is also a backlash about what has happened to the Congressional process. In my opinion the government has picked so many fights that it will be hard to continue to prop up the banking industry. Banking is recording huge profits. It would be a tough sell to say that we need to infuse another X billion of dollars. Look at what happened to the jobs bill. Health Reform is on the mat today. Even if it passes or it doesn’t Congress is getting the lowest approval rating ever.

    So though I share the concern……

  32. 32

    […] 10, 2010 · Leave a Comment Seattle Bubble carries a nice interactive viz showing that while median price is still falling all over Puget […]

  33. 33
    3rd Generation says:

    I apologize for my ignorance but I’m always trying to learn: WTH is an IOW as used in reply 1 from softwarengineer » Mar 9, 2010 at 1:55 pm ?

    Cheers! and thanks.

  34. 34
  35. 35
  36. 36

    […] our discussion about the wide variety of median price changes in neighborhoods around King County, a question came up: How can two thirds of the neighborhoods be experiencing median price declines but the county-wide […]

  37. 37
    Lurker says:

    RE: Hugh Dominic @ 26

    I would guess that the current administration would do whatever possible to prop it up until they can make it through re-election.

    @buystocks – wife and I went to some open houses a few weeks back. We weren’t the only ones looking but it did feel very quiet. One agent looked genuinely hurt when I gave her back the info sheet when we left. It was very nice, colorful and probably wasn’t cheap to print, I was just trying to save them a little money.

  38. 38
    3rd Generation says:

    Ira, thanks. John.

  39. 39
    Shameer says:

    RE: The Tim @ 9
    There is actually a name for this effect. It’s called Simpson’s Paradox

  40. 40
    ARDELL says:

    Thank you Shameer,

    I find the smaller geographic samplings are more relevant, but given low volume relative to 2006 and 2007, a month of sales YOY doesn’t give enough data to form a valid conclusion.

    I end up having to expand the number of days from 30 to at least 90, when evaluating smaller geographic samplings, especially on the Eastside. It’s also why I never do stats for places like Medina. I get reports from Title Companies saying “volume doubled” and then I look to see they had 2 sales instead of 1 :)

  41. 41
    Daniel says:

    Disclaimer: I was not aware of Tims new post until now. I submit this anyways.

    RE: Shameer @ 39 – Lets create a (for now) fictitious example that would lead to the observed behavior: a stable median while most areas seem to suggest a decline.

    The simplest way this could be achieved is by increasing sales volume in high-priced areas and a decreasing or constant sales volume in low-priced areas. Even when the median just considering the high-priced areas or low priced areas alone is falling, the overall median would remain stable.

    Such a scenario seems not implausible, as people that still hold their jobs and whose resources have not reduced might just consider to upgrade to a home in a better neighbourhood.

    Of course I still am not convinced that what we see is more than statistical noise.

  42. 42
    The Tim says:

    By ARDELL @ 18:

    One of the things I have tried is to segregate the majority of sales by price and then check that as to up or down. Do you think that improves accuracy?

    You’ll have to elaborate a bit on that, as I’m not sure exactly what you’re saying. Feel free to drop me an email if you’d like to discuss methods a little more in depth.

    RE: Daniel @ 28 – You sound like you have a pretty strong background in statistics. Care to drop me a line to discuss your ideas via email?

    RE: Shameer @ 39 – Nice, thanks for the link. You learn something new every day.

    RE: Daniel @ 41 – Hey, good guess! :^)

  43. 43
    ARDELL says:

    For instance, I have first identified the majority of sales being less than X amount in a 12 month period, which as I recall was in the $650,000 range in a year’s time in a given area. Once I determine that say 87% of all sales were at or below that number (for some reason it often comes out to 87%) then I run the “up or down” stats on sales that fall into the $650,000 or less category, without including the 13% of sales that are over $650,000. This way one month with an inordinate number of $650,000 plus sales does not skew the data.

    Does that make any sense to you, or might you have another suggestion? RE: The Tim @ 42

  44. 44
    David Losh says:

    RE: Shameer @ 39

    Exactly. Thank you.

  45. 45
    Daniel says:

    By ARDELL @ 43:

    For instance, I have first identified the majority of sales being less than X amount in a 12 month period, which as I recall was in the $650,000 range in a year’s time in a given area. Once I determine that say 87% of all sales were at or below that number (for some reason it often comes out to 87%) then I run the “up or down” stats on sales that fall into the $650,000 or less category, without including the 13% of sales that are over $650,000. This way one month with an inordinate number of $650,000 plus sales does not skew the data.

    This should not be necessary if you use a median (as opposed to the arithmetic mean). While the arithmetic mean minimizes the sum of differences squared, a median minimizes the average of absolute deviations. While the median is more appropriate for a skewed distribution the mean is in a certain sense “optimal” (no other unbiased estimator can have a smaller variance) for a normal distribution. By cutting out the large price tail of the distribution, you will significantly influence the mean but not the median.

    Such a cut can be useful in other situations though, for example when there is a large unwanted background that you are not interested in, and the part of the data sample containing what you are interested in is located in a much smaller region. In the case where you are interested in the whole sample all you do is bias your result.

  46. 46

    […] There is a very interesting post and conversation about this over at Seattle Bubble. I encourage you to take a look. […]

  47. 47
    Choc Donut says:

    Umm, I’m looking at the MLS every week, and prices are down a lot with houses sitting in south seattle (less desireable). While supply is low on QA, what’s interesting is that across town, from Ballard to tony Magnolia and Montlake, top prices are coming down fast. You can get in these hoods for 450-500k, when three years ago, you couldnt find anything real under 700k. THis market is soft, a bunch of people are predicting what the charts show, an imminent explosion of alt-A mortgages. Plus boomers are gonna be in full panic mode when they discover their retirements aren’t going to be covered after all, and they better get out now. Housing won’t be worth buying until after that apocalypse, and this country’s next generation are so mal-raised and mal-educated, and our leadership so corrupt, you might as well just take the money and run, as in expatriate.

  48. 48

    […] th {text-align: center;} .top_row {font-weight: bold;} .CNNTable img {border:0;margin:0;}With some neighborhoods around Seattle showing year-over-year gains in their median prices, it would be informative to see just what kind of houses people are buying at the median price […]

  49. 49

    […] let’s have another look at the viz of neighborhood YOY median price changes that we introduced in March. King County Neighborhood Prices Powered by TableauNot too surprisingly, median prices were down […]

Leave a Reply

Use your email address to sign up with Gravatar for a custom avatar.
Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Please read the rules before posting a comment.